Cadence: multiple concurrent proposers, and a huge step toward MEV resistance

Cadence is a new multiple concurrent proposer (MCP) consensus protocol by Category Labs. When fully implemented and released on Monad, it will support 100-millisecond blocks and 200-millisecond finality, and will significantly enhance Monad's resilience to MEV attacks that affect users of other blockchains.

Cadence matches the optimal good-case latency of single-leader consensus while supporting much shorter block intervals, and it removes the single leader's monopoly over transaction ordering.

Paired with BTX, Category Labs' design for an encrypted mempool, it forms a major step toward solving MEV.

There is a full interactive explainer and an academic paper, Cadence: Extreme Pipelining with Multiple Concurrent Proposers. This post is a short summary of what was announced and why it matters.

The two limits of a single leader

Most modern BFT consensus protocols, including MonadBFT, finalize blocks in a chain: each block builds on its predecessor, so a new block cannot be proposed until the previous one has been produced and propagated across the network. That design ties the block interval to network latency — you cannot go much faster than the time it takes a block to travel the network, no matter how much spare capacity validators have.

A single leader per block also concentrates power. The leader alone decides which transactions are included and in what order, which is a censorship vector and the root of most maximal extractable value (MEV): the leader, or arbitrageurs collaborating with the leader, can insert, reorder, or drop transactions to extract value from users.

How Cadence works

Cadence attacks both limits at once.

Extreme pipelining. Cadence divides time into equally spaced slots, one block per slot, with each block finalized in its own independent consensus instance. Crucially, blocks do not build directly on their predecessor, so instances run in parallel and none waits for an earlier block to finish or propagate. This decouples the block interval from network latency — the network can open new slots as fast as it likes while each finalizes on its own schedule.

Multiple concurrent proposers (MCP). Instead of a single leader, several validators propose for each block. Under synchrony, Cadence guarantees that a transaction a correct proposer includes cannot be censored or deferred (short-term censorship resistance), and that no proposer can craft its proposal in reaction to the others' (hiding).

To make this work, Category Labs introduces a general framework that turns any one-shot consensus meeting its slot-consensus specification into a full multi-shot protocol, and instantiates it with two components:

Chorus — the per-slot consensus, whose fast path finalizes a block in an optimal three rounds, with speculative finality one round earlier.
Conductor — an orchestrator that opens slots at an even cadence, slowing down under asynchrony to keep the number of open slots bounded.

To the authors' knowledge, Cadence is the first MCP protocol to provide short-term censorship resistance and hiding at the fast-path latency of single-leader consensus. The paper proves safety, liveness, censorship resistance, and hiding under partial synchrony with optimal resilience (n = 3f + 1).

The numbers

In simulation over Monad's 200 validators with five proposers per slot, the paper reports:

~219 ms average finalization (167 ms to speculative finality)
100 ms block interval, with a transaction waiting on average just 50 ms to enter a proposal

A step toward MEV resistance

Censorship resistance and hiding are what make Cadence more than a latency win. Because no proposer sees the others' contents in time to react, and because a correct proposer's transactions cannot be dropped, each block proposal becomes much more resilient to MEV attacks from other proposers.

Cadence closes the reordering half of the problem. The other half — seeing transaction contents before they are ordered — is closed by an encrypted mempool. BTX provides a threshold encryption scheme that makes encrypted mempools pragmatic. As a result, transactions stay encrypted until they are committed, so no one can read them in time to front-run or sandwich. Together, Cadence and BTX are designed to attack MEV from both directions at once.

Monad's work on asynchronous (deferred) execution — first pursued for raw performance — is what makes these protocol-level advances tractable, now bearing fruit in MCP and encrypted mempools.

This signals the near-future direction of Monad: becoming the blockchain that delivers MEV resistance, while also improving on industry-leading performance.

Learn more

About Monad

Monad is a high-performance, EVM-compatible Layer 1 blockchain built to power the financial layer of the internet. Fully EVM-compatible, Monad delivers 10,000 TPS, 400ms block times, 800ms finality, and near-zero fees, without requiring specialized hardware. The network runs on consumer-grade machines, supporting accessible participation and decentralized network operation: over 200 independently operated validators across 30+ countries and 55+ cities secure the chain today.